|
Abstract:
|
Doubly censored data are data that are subject to both left and right censoring. They are common in longitudinal studies when the event of interest might not be observed because it either has occurred at an unknown time before the subject entered the study or has not occurred yet at the end of the study. In this talk, we consider doubly-censored data from familial-type studies, where observations from the same family are correlated. This results in clustered survival data, which we model by a class of linear transformation models. The transformation models expand the horizon of survival models, as they include both the proportional hazards and proportional odds models as special cases, but they pose considerable computational challenges to likelihood approaches for doubly-censored clustered data. An effective EM algorithm is developed to overcome the computational difficulties and it leads to stable nonparametric maximum likelihood estimates (NPMLEs). We investigate the asymptotic properties of the NPMLEs and demonstrate the semiparametric efficiency of the finite-dimensional parameters. In addition, a computationally efficient method is proposed to estimate the standard error
|
